Safety of paracetamol in osteoarthritis

Background Pain is one of the most troubling symptoms that warrant people with chronic conditions such as osteoarthritis (OA) to seek pharmacological treatment options. Paracetamol is currently the first-line drug recommended by most practice guidelines for managing pain in OA due to its perceived safety profile over other analgesics. Despite its modest effectiveness, the usage of paracetamol is widespread either as a prescription medication or over the counter drug for acute and chronic painful conditions. Nevertheless, the adverse effects have been overlooked in its long-term therapeutic usage. There is a growing controversy, highlighted in the National Institute for Health and Care Excellence OA (NICE) 2014 guidance on OA, that paracetamol is not as safe as thought previously, especially at the highest therapeutic dose of 4gm/day for the long term. Additionally, based on a limited number of studies, the association of gastrointestinal, cardiovascular, and renal events and all-cause mortality with the use of paracetamol is still unclear. Gaining knowledge of these adverse effects would help address this gap in the evidence and provide insights for clinical decision-making. Eventually, it might inform modification of the current recommendations and guidelines for chronic painful conditions such as OA and help clinicians discuss these side effects with patients when considering treatment with paracetamol. Aim To examine the long-term safety of paracetamol in people with OA and the general population. Objectives Objective 1: To systematically review the research evidence, including randomised controlled trials (RCTs) in people with OA and observational studies (irrespective of any specific underlying condition) on adverse effects of paracetamol. Objective 2: To undertake a propensity-score matched cohort study to confirm the associations identified from the systematic review and examine the incidence of perforation ulcer bleeding (PUB), uncomplicated ulcers, lower GI bleeding, heart failure, myocardial infarction, hypertension, and chronic renal failure in those exposed to paracetamol in the general population compared to unexposed people using electronic medical records of patients registered in UK general practices. Also, to examine the dose-response relationship between paracetamol and related adverse events using real-world data in the general population. Methods 1) A systematic literature search and meta-analysis of published RCTs and observational studies For systematic review and meta-analysis (SR/MA), the databases: MEDLINE, PUBMED, EMBASE, AMED, CINAHL, Web of Science, and Google Scholar were systematically searched for published literature in any language to the end of November 2018 for (1) RCTs of paracetamol in symptomatic and radiographic OA, and (2) observational studies for the treatment of any underlying conditions, to assess the risk of gastrointestinal, liver, renal and cardiovascular adverse events. The SR/MA included studies investigating oral paracetamol in people aged ≥18 years and reporting adverse effects. RCTs and cohort studies were estimated using the risk ratio (RR) or hazard ratio (HR), while case-control studies used the odds ratio (OR) and 95% confidence interval (CI). The results were pooled based on the random-effects model, as appropriate. The risk of bias was assessed using the modified Cochrane risk of bias tool for RCTs and the modified Newcastle Ottawa scale for observational studies, respectively. 2) A propensity-score matched cohort study using the Clinical Practice Datalink (CPRD) Study participants were 65 years old and over registered with up-to-date practices for at least 12 months from January 1998 to July 2018. Participants who received at least two paracetamol prescriptions from their general practitioner (GP) within six months were defined as exposed to paracetamol. The first prescription date of these two prescriptions was the index date. Paracetamol non-exposure was defined as participants who did not receive two paracetamol prescriptions within six months over the study period. They were matched with the exposure by age, gender, and propensity score and assigned the same index date as their matched exposure. Logistic regression was used to calculate the propensity score. Follow-up started 12 months after the index date (i.e., the landmark date) to minimise immortal-time bias and provide an exposure window between the index date and landmark date to accumulate several prescriptions and doses for dose-response analysis. Participants were followed until the first incidence of the outcome of interest, deregistration from the practice, death, or end of the study, whichever came first. Kaplan-Meier (Nelson-Aelon) curves were used to estimate the cumulative probabilities of the outcome. The HR and 95% confidence interval (CI) were calculated using the Cox-proportional hazards model. Multiple imputations (MI) were conducted to account for missing data, followed by the inverse probability treatment weighted method (IPTW) to balance the covariates. Results 1) SR/MA of published RCTs in people with OA and observational studies without any underlying condition There were 3,722 citations (1,997 RCTs and 1,725 observational studies) for titles and abstracts screening. After reviewing the full text, 24 RCTs, 16 cohort studies, and 26 case-control studies met the inclusion criteria. GI risk In 24 RCTs (7,299 participants), paracetamol was associated with increased incidence of any treatment-related adverse events (RR 1.37; 95%CI 1.06, 1.77), diarrhoea (RR 2.12; 95%CI 1.36, 3.32) and liver function complications (RR 3.99; 95%CI 2.05, 7.77) compared to placebo. In 12 age and gender-matched case-control studies (7,894 participants), the pooled OR associated with paracetamol for upper GI bleeding was 1.34 (95%CI 1.10, 1.63). A cohort study including 382,404 participants reported a dose-response relationship between paracetamol doses and perforation ulcer bleeding (PUB). The RR was 1.11 (95%CI 1.04, 1.21) with low dose paracetamol (measured as medication possession rate) and 1.49 (95%CI 1.34, 1.66) with very high dose. CV risk In two cohort studies (137,955 participants), there was an indication of the association between paracetamol exposure and CV risk, as the pooled RR showed a dose-response relationship with RR of 1.11 (95% CI 1.00, 1.22) when paracetamol was used for 1-4 days/month to 1.52 (0.92, 2.51) with exposure >22 days/month, but not from two RCTs (1,011 participants) (RR 1.05; 95%CI 0.07 to 15.38) and three case-control studies (42,180 participants) (OR 0.93; 95%CI 0.80 to 1.09). Renal risk The data for renal adverse effects was insufficient in RCTs. However, one cohort study with 19,163 participants reported a dose-response relationship between paracetamol exposed and end-stage kidney disease compared to unexposed. Three case-control studies also suggested an increased risk for end-stage kidney disease with the high therapeutic dose of paracetamol compared to unexposed. Two cohort studies suggested a dose-response relationship between paracetamol exposure and chronic renal failure. 2) A propensity-score matched cohort study using the Clinical Practice Datalink (CPRD) GI risk From 1998 to 2018, 180,483 paracetamol users were identified from the database. They were successfully matched with 402,487 non-users by age, sex, and practice. There were 158,048 successfully PS-matched, i.e., 79,024 paracetamol users and 79,024 non-users. Paracetamol was associated with an increased risk of PUB (HR 1.24; 95%CI 1.16, 1.34), uncomplicated ulcer (HR 1.20; 95%CI 1.10, 1.31), and lower GI bleeding (HR 1.36; 95%CI 1.29, 1.46). A dose-response was observed for PUB and uncomplicated ulcers with significant P for the trend. CV risk Paracetamol exposure was associated with an increased risk of heart failure (HR 1.09; 95%CI 1.06, 1.13) and hypertension (HR 1.07; 95%CI 1.04, 1.11), not for myocardial infarction (HR 0.99; 95%CI 0.94, 1.04). No dose-response was observed for any CV outcome in the present cohort study. Renal risk Paracetamol exposure was associated with an increased risk of chronic renal failure (HR 1.19; 95%CI 1.13, 1.24) compared to unexposed in the present cohort study. A dose-response trend was also apparent, with a significant P for the trend for chronic renal failure. Conclusion Paracetamol is associated with diarrhoea and abnormal liver function from RCTs evidence. It is also associated with an increased risk of gastrointestinal, cardiovascular, and renal adverse events from published observational studies. The population-based propensity score-matched cohort study of Chapter 3 in this thesis has confirmed most of these adverse effects, where the likelihood of confounding by indication has been minimised. Thus, adverse events identified through this research should be carefully weighed against the potential benefits of paracetamol. Notably, in conjunction with published paracetamol efficacy data for individual patients, while deciding to prescribe paracetamol, especially for older people or people with osteoarthritis and other chronic painful conditions. Further study on other outcomes, such as all-cause mortality, has yet to be confirmed